Drone as a tool for coastal flood monitoring in the Volta Delta, Ghana

Delta environment is dynamic, highly vulnerable and challenged due to adverse impacts of flooding probably as a result of relative sea level rise, extreme oceanic wave actions, and subsidence or anthropogenic impact such as dam construction etc. (Ericson et al. 2006; Wong et al. 2014; Appeaning Addo 2015). Variety of methods that differ in approach, accuracy, cost and duration have been developed to meet the demand for higher accuracy in delta environment monitoring (Appeaning Addo et al. 2008). These methods can broadly be categorized as in-situ (physical surveying) method and remote sensing approach.

The physical surveying method has resulted in the production of historical maps at varying scales, which can be combined with recent information on the delta environment to detect and measure change in the shoreline morphology (Appeaning Addo et al. 2008). The remote sensing approach includes aerial photographs, satellite imagery, airborne light detection and ranging technology (LIDAR), and video technique.

The land based methods and the remote sensing approaches have their strengths and weaknesses, which can affect their levels of accuracy, the reliability of the output maps generated and their usability. These strengths and weaknesses have been discussed extensively in the literature (see Basha et al. 2008; Kussul et al. 2008; Matgen et al. 2011;Turner et al., 2016). Selecting a particular method for monitoring is therefore influenced by several factors such as the type of feature to be monitored, the accuracy required, availability of funding, type of output desired, the intended use of the results and geographical location (Dolan et al. 1991; Klemas 2015).

Physical surveying generated historic maps detailed the mapped position of physical features influenced by flooding overtime as estimated on site by a land surveyor. Such maps provide accurate information for change detection as the surveyor is physically present during the data collection period (Graham et al. 2003; Leatherman et al. 2005). However, the approach tends to be limited both spatially and temporally due to several factors (Morton 1991; Smith and Jackson 1992). The method is time consuming, expensive and impractical for long distance mapping (Leatherman et al. 2005). It is also associated with human errors that can affect its reliability (Appeaning Addo et al. 2008). Aerial photographs are the most commonly used data source in coastal environment monitoring (Moore 2000). In addition to having the ability for stereo pair mapping, they also provide good spatial coverage (Lane et al. 2001). However, temporal coverage is very site specific and the process of acquiring the near vertical photographs as well as the laborious field preparation can be expensive (Whitehead et al. 2014; Darwin et al. 2013). This short coming affects the repeatability of the monitoring process.

Satellite imagery approach covers large area and provides detailed spectral information (Boak and Turner 2005) as well as very high resolution (< 1 m) satellite data (Ford 2013).Satellites have short revisit time, provide multi-spectral data and have the capability for stereo pair mapping. However, they are affected by cloud coverage, dust during image acquisition, pixel resolution and operational cost (Whitehead et al. 2014; Darwin et al. 2013).

LIDAR has the ability to cover hundreds of kilometers of coastal area in a relatively short period and produce extremely dense and accurate elevation measurements (Stockdon et al. 2002). It is limited in its temporal and spatial availability because of cost (Boak and Turner 2005). Digital video imaging also has the capability to monitor detailed changes in the coastal systems (Boak and Turner 2005) and record flooding events in real-time. It also has the ability to collect both time-averaged and instantaneous images (Angnuureng et al. 2016). The sensor has a fixed location and a fixed area of coverage that makes it unsuitable for wider coverage monitoring.

Recent developments have resulted in the use of Unmanned Aerial Vehicles (UAVs) also known as drones in delta environment monitoring (Mancini et al. 2013; Klemas 2015). This is due to the relatively low cost of operation that allows for frequent missions, increased spatial coverage, no required installation points, rapid deployment, better quality outputs (Vousdoukas et al. 2011) and as a tool for effective communication. The ability of drones to perform missions and acquire data autonomously as well as its maneuverability capacity helps in complex coastal environmental disaster monitoring (Whitehead and Hugenholtz 2014; Wallace et al. 2012). Datasets produced by drone have high resolution (< 10 cm) (Harwin and Lucieer 2012) and supports the development of high resolution digital elevation models (DEMs) that facilitate change detection and measurement (Hugenholtz et al. 2013). Drone based monitoring of disaster events has significant advantages due to its timeliness, event based rapid availability and better quality/resolution (Pérez-Alberti and Trenhaile 2015). Drones are also being used in advocacy to booster community awareness in various environmental hazards, including deforestation and bush fires and as a communication tool in stakeholder engagement at various levels (Goldberg et al. 2013; Mohammed et al. 2014). Appeaning Addo (2016) used drone video recordings to communicate erosion dynamics and flood problems to policy makers in Ghana (parliamentarian and local authorities), which influenced their perception about these problems (Owens 2016). The use of drone is therefore the best way in data gathering to arm decision makers and local authorities with more accurate picture of environmental impacts (STAFF AG 2017).

A major challenge to the operation of drone in the coastal environment of delta regions is the presence of strong winds (Gonçalves and Henriques 2015). Flights are therefore limited to specific times to reduce the impact of strong winds. Other challenges include platform instability, view angle, data processing tools and short flight times due to battery constraints (Elaksher et al. 2017). This paper presents the application of drones in flood monitoring in the Volta Delta in Ghana and sets up agenda for further studies in the Volta Delta. The drone technology was adopted for this study due to its advantages over the other methods, which will ensure repeated flood monitoring activity in the Volta Delta environment.

Flooding frequency and coastal erosion intensity as a result of intensive rainfall, oceanographic conditions (waves, sea level rise and tides) and human activities (watershed management) has increased in frequency and impact in the Volta Delta significantly. Construction of the Akosombo dam on the Volta River has been identified as a major cause of erosion and flooding problems in the Volta Delta region. Runoff before dam construction was higher (87.5 mm/yr) and more varied than the post-dam period with value of 73.5 mm/yr. (Oguntunde et al. 2006). The dam construction (completed in 1965) has resulted in controlled water flow, which has changed the natural flooding of the area (Corcoran et al. 2007), although periodic high flooding from the tidal waves persist. The yearly sediment transport before the dam construction was about 7.5 million m³/s (Bollen et al. 2011). Since the construction of the dam, there are no peaks in flow discharge and the sediment transport is reduced to only a fraction of the original transport (Bollen et al. 2011). It is estimated that coastal erosion and flooding have destroyed about 5000 houses since the 1960s, displaced households, destroyed sources of livelihoods and forced migration within and out of the Delta (Boateng 2009; Karley 2009; Lumor 2015). Though riverine flooding occurs and particularly during the dam spillage, tidal flooding is dominant along the coastline of the Delta. The last reported Dam spillage was in November 2010 (Owusu and Waylen 2013) which cause flooding in communities along the river. However, tidal flooding in the Delta is on the increase as observed by previous reports (Bokpe 2010; Fagotto 2016).

Study area

The Volta Delta is located in the eastern coast of Ghana within latitudes 5⁰25′ and 6⁰20’ North and longitude 0⁰40′ and 1⁰10′ East (Fig. 1). The delta is defined as the land below the 5 m contour in the lower portion of the Volta river basin within the Accra-Ho-Keta Plains (Appeaning Addo et al. 2018). Nine administrative districts that are responsible for managing resources in the Delta region are located wholly or partially in the Volta Delta. The delta is characterised by a fairly uniform, moderately steep shoreface with a gradient of between 1:120 and 1:150 down to 15 m, which is considered as the close-out depth for significant wave-induced sediment movement on this coast (Rossi 1989; Anthony 2015). The wave climate in the region consists of swell waves from south-west direction (Angnuureng et al. 2013). The average significant wave height is about 1.4 m (Wellens-Mensah et al. 2002) while the annual maximum are around 2.5 – 3 m (Roest 2018). Some variability in the wave climate is present over the year. The highest waves occur from July to August, while the lowest waves occur around January and February (Roest 2018). The tidal regime is semi-diurnal with a tidal range of about 1 m (Wellens-Mensah et al. 2002). The tidal currents are weak and their impact on the shoreline morphology is limited (Wellens-Mensah et al. 2002). Although the environment is microtidal, a study by Angnuureng et al. (2016) identified that short term evolution of the shoreline is affected by tidal cycles from neap to spring. Presently, the sea level is rising at a rate of about 3.1 mm/yr. (Sagoe-Addy and Appeaning Addo 2013). The sea level is predicted to continue rising in conformity to the global trend (Armah et al. 2005).

The basic winds along the study area are southwest monsoon. It blows from the south-west direction (210°-240°) from the sea to land at about 45° angle to the coast and it is approximately in the same direction with the waves (Angnuureng et al. 2013). During the Harmattan season (December–February) winds occasionally blow from the northwest. The monthly average wind speed ranges between 1.7 and 2.6 m/s (Angnuureng et al. 2013).

Several studies have identified the Volta Delta as coastal erosion and flooding hotspot (Boateng 2012; Jayson-Quashigah et al. 2013; Appeaning Addo 2015). One community that suffered huge destruction of properties and displacement of households as a result of coastal flooding and erosion is Fuvemeh, a fishing community with over 1,500 inhabitants (see Fig. 1). The problems associated with the flooding incidents in the community require scientifically measured and reliable information to aid in effective management and developing a resilient community.

Rainfall data was obtained from Ghana Meteorological Agency (GMet) in Accra. Data on impact of flooding on people and buildings in the Keta environ within the Volta Delta region was obtained from the Ghana National Disaster Management Organisation (NADMO) in Keta. The data were used to analyse the rainfall trend in the region, frequency of human displacement and building collapsing as a result of flooding in the delta region. The rainfall data included recorded rainfall obtained from rain gauge station in Akatsi in the delta region from 1986 to 2015. The recorded number of people displaced and buildings collapsed by flooding were between the periods 2008 and 2014.

Orthophoto map of 2005 (0.5 m ground resolution) and satellite imagery of 2014 (1.0 m ground resolution) were obtained from the Survey and Mapping Division in Ghana and Digital Globe Foundation respectively for the Fuvemeh community. A DJI Phantom 3 drone was used to obtain aerial photographs (0.06 m ground resolution) of Fuvemeh community in February 2016 with repeated surveys in August 2016 and June 2017. Appeaning Addo et al. (2008) validated the 2005 orthophoto map and concluded that the map is accurate and reliable. The orthophoto map was therefore adopted as reference and used to validate the satellite imageries and the drone aerial photographs by checking the positional accuracies of selected features (5 buildings). The coordinates of the features on the satellite imageries and the drone pictures matched their conjugate coordinates on the orthophoto maps. This gave confidence in using the data for the study.

The shoreline positions, represented by the High Water Line (HWL) as proxy were mapped and appended in GIS environment to detect and measure change within the period under study. The Digital Shoreline Analysis System (DSAS) software developed by Thieler et al. (2017) was used to analyse the shoreline changing trends along equally spaced perpendicular transects along the coast. The transect intervals were set at 50 m to enable more areas to be covered in the rates of change estimation. The average rates of erosion were statistically estimated using the end point rate of change method (EPR). The EPR was adopted because it is the most commonly used method to compute shoreline rate of change (Genz et al. 2007). The method is simple and requires only two shoreline positions to obtain a rate of change. It calculates the rate of change by dividing the distance of shoreline movement by the time elapsed between the earliest and the latest measurement, which can be the oldest and the most recent shoreline positions (Crowell et al. 2005).

Analysis of data obtained from NADMO revealed that several areas in the delta region are experiencing flooding problems. In all, 28 flooding events were recorded at varying intensity, which displaced over 2,000 people and destroyed about 306 buildings (Fig. 2).

It can be seen from Fig. 2 that between 2008 and 2014, several people were displaced by flooding event and significant number of houses were also destroyed. The highest number of displaced people (685) occurred in 2011, while the lowest number occurred in 2013. The highest number of houses (90) were destroyed in 2008 and the lowest number of houses (5) were lost in 2013. The high destruction of houses in 2008 can be attributed to storm surge which flooded the coastal zone in the delta region (Ghanaweb 2008) and flooding from rainfall. The severe flood experienced in 2011 was due to heavy rains that caused rivers to overflow their banks (Duodu 2011). Analysis of the rain gauge recorded data is presented in Fig. 3, which shows the annual rainfall distribution pattern and the general trend in Akatsi as per the linear trend line. It can be seen that between 1986 and 2015 the area recorded substantial amount of rainfall that resulted in flooding in some areas. The graph in Fig. 3 shows that 1986 to 1995 had mean rainfall value of 780.76 mm, from 1996 to 2015 the rainfall value was 960.17 mm and then from 2006 to 2015 the mean rainfall average was 818.67 mm. There was increasing trend from the first decade to the second decade of 1996 to 2005 and the trend decreased from the second decade of 1996 to 2005 and the third decade of 2006 to 2015. The general trend indicates increasing rainfall pattern in the delta region. Comparing Figs. 3 and 4 from 2008 and 2014, which falls within the 3rd decade, the decadal trend reduced similar to the significant reduced building destruction and displaced people, inferring reduced flooding events.

Analysis of the shoreline change using the 2005 orthophotos and 2017 drone images shows that the shoreline has moved inland several meters (about 100 m in some instances) along some perpendicular transects, while along some transects land has been gained (about 80 m in some instance) as can be seen in Fig. 4.

The shoreline migration trend analysis also revealed that out of a total area of about 375,229 m², about 138,118.239 m² of it has been lost as a result of erosion, which represent about 37% of the total land area. A total of about 77 houses have been destroyed by erosion between 2005 and 2017, which is about 42% of the total houses in the Fuvemeh community. The destruction of the houses have resulted in the displacement of over 300 inhabitants in the Fuvemeh community. Figure 5 shows houses that have been lost between 2005 and 2017. Strong erosion as result of natural processes (e.g. sea level rise, storm surge, energetic swell waves, subsidence, etc.) and anthropogenic activities (dam construction, sand mining, over harvesting of mangroves, etc.) has damaged properties and destroyed the environment. It has also destroyed natural fish landing sites, which has affected the local fishing business (the main source of livelihood) and rendered some of the inhabitants’ jobless (Boateng 2012).

Figure 6 shows part of the community in August 2013 from satellite imagery and the same area as captured by the drone in February 2016. The portion labelled A is the location of the only school in the community on both the August 2013 and February 2016 imageries. Comparing the two imageries in Fig. 6, it can be seen that the position of the school from the shoreline has changed. The portion labelled B on the satellite imagery shows part of the community in August 2013 which is now flooded on the February 2016 image. The lagoon on the August 2013 image is also lost on the February 2016 image.

Figure 7 shows the remains of the school building in May 2016 (A) and June 2016 (B) respectively.

Figure 8 shows drone image of the flooding event in the community captured on 9th February 2016. The energetic wave action, which were larger than the usual waves in the area, was aided by high tides and the prevailing relatively low topography, which is below 2 m above sea level (Boateng 2009) to break closer to the community and thus flood the Fuvemeh community. Figure 9 shows the devastating effect of the flood water on dwellings places (building labelled A) and sources of livelihood (labelled B).

The swash of the waves deposit debris onto the beaches, while the strong backwash carries properties of the inhabitants into the sea. The inhabitants have to salvage their properties and carry them to safe places (Fig. 10). Occupants of such flooded houses abandon the buildings and either stay with friends in safer areas or leave the community to other places as there is no government intervention. Some of the affected households remove the roofing sheets of their buildings to put up new buildings in less vulnerable areas.

Analysis of the rainfall data revealed that the general pattern in the rainfall variability shows an increasing trend, which results in flooding and erosion. According to Allotey et al. (2008), the soil is often flooded during the wet season. Other factors that cause flooding along the coast in Fuvemeh and the Volta Delta include the occasional overflow of the Volta River when the spillways of Bagre and Akosombo dams are opened amid heavy rains (Biney 2010; Boateng 2012), strong energetic wave activities from the sea (Oteng-Ababio et al. 2011), occasional storm surge which has increased in frequency and intensity in recent times (Gakpo 2016), increasing relative sea level rise as a result of climate change, possible increased subsidence due to ground water extraction for irrigation and to a lesser extent tidal actions. Although the rate of subsidence has not been measured, it is expected to probably be between 1 and 2 mm/yr. based on other deltas (Syvitski 2008; Appeaning Addo et al. 2018). The rising sea level coupled with the relatively deep nearshore bathymetry (Jayson-Quashigah et al. 2013; Sagoe-Addy and Appeaning Addo 2013) enables the strong waves to break closer to the shoreline. Any sudden increase in wind intensity may result in a surge, which will further push the breaking waves more inland to flood the low lying areas. Over harvesting of mangroves for domestic use and converting mangrove vegetated areas to fish farms have reduced the natural protection they offer and thus rendered the environment vulnerable.

Analysis of the NADMO data shows the impact of flooding on the environment and the people. Flooding therefore has serious socioeconomic implication for the Fuvemeh community. The destruction of the school in the community will have huge implications on formal education delivery in the area. It emerged from informal discussions with members of the community that the destruction of the school building has increased the difficulty in assessing education facilities as the displaced school children now have to travel to nearby communities to attend school. The school building, according to some members of the community interviewed during the study, is used as a place of refuge during flood event in the community and the surrounding villages in the past. The collapse of the school building will affect rescuing effort of NADMO during such disasters. The situation has also forced some of the inhabitants to migrate from the community to nearby communities, while others have moved from the delta region to other parts of the country and beyond (Mumuni et al. 2017). Destruction of natural fish landing sites by erosion has collapsed the small scale fishing industry and increased poverty in the community. This is expected to have a huge impact on the protein intake of the inhabitants in the community and the surrounding communities as well the food security in the country.

The changes in the shoreline migration trend being experienced in the study area are due to several factors such as energetic wave action, storm surge, dam constructed on the Volta River in Akosombo, the low topography of the beach system and sea level rise (Sagoe-Addy and Appeaning Addo 2013, Giardino et al. 2018). The construction of coastal defense structures in Ada (the up-drift side of Fuvemeh) and the reduced sediment discharge from the Volta River (Boateng 2012) has affected the sediment regime in the area. The defense structures are made up of groynes, which traps sediment and thereby starve the down-drift side of sediment as observed by Angnuureng et al. (2013) in Keta. The action of the groynes further enhances the impact of the reduced sediment discharge from the Volta River which has resulted in the observed morphological changes.

Coastal flooding and erosion are major issues in Fuvemeh and the entire Volta Delta. Their impact has intensified as a result of increased rainfall that results in rivers overflowing their banks; increased oceanographic conditions such as frequent storm surge activities, energetic swell waves, tidal activities and sea level rise that impact vulnerable communities like Fuvemeh; upstream water catchment management such as construction of Akosombo dam on the Volta River which has influenced the sediment budget regime in delta system as well as the nearshore; overharvesting of mangroves that has destroyed the natural defence the vegetation provides and thereby opened the area to intense wave and tidal attacks; and sand mining for construction that has further affected the sediment budget regime. Coastal erosion and flooding in Fuvemeh have collapsed buildings (about 42% of the total houses); displaced inhabitants (about 300 inhabitants); destroyed sources of livelihood (about 37% of Fuvemeh land, mainly natural fish landing sites along the coast, has been eroded and this has affected the local fishing business); and increased migration in the vulnerable areas (affected social and cultural dynamics in the Fuvemeh community). The problems will increase in future under changing climatic conditions. There is therefore the need to develop a sustainable adaptation approach to effectively manage the situation and increase the resilience of the inhabitants. Although the government has adopted building hard engineering structures to manage erosion and flooding, not all the areas experiencing such hazards can be managed using this approach. It is suggested that relocating the Fuvemeh community to a more secured place will be more appropriate and cost effective. Alternatively, a soft engineering approach such as beach nourishment can be adopted to augment the sediment shortage in the area. The study has further demonstrated the effectiveness of using the drone technology in monitoring flood and erosion in a delta community. The technology can therefore be applied in the fields of natural hazards, disaster response and high resolution terrain analysis.